[petsc-users] Failure to configure superlu_dist

[Satish Balay]

Glad you were able to figure this out.

So having an incompatible MPI_ROOT env variable set can break builds [breaks cmake or mpif90?]

Satish

On Mon, 28 Mar 2022, Alexander Lindsay wrote:

> Ok, nothing to see here ... This was user error. I had MPI_ROOT set to a
> different MPI install than that corresponding to the mpi in my PATH.
> 
> On Mon, Mar 28, 2022 at 2:13 PM Satish Balay <balay at mcs.anl.gov> wrote:
> 
> > I think moose is bound now to a petsc-3.16.5+ snapshot
> >
> > >>>>>>
> > -- Could NOT find MPI_C (missing: MPI_C_HEADER_DIR) (found version "4.0")
> > -- Found MPI_CXX: /home/lindad/mpich/installed/lib/libmpicxx.so (found
> > version "4.0")
> > -- Found MPI_Fortran: /home/lindad/mpich/installed/lib/libmpifort.so
> > (found version "3.1")
> > -- Configuring incomplete, errors occurred!
> > <<<<
> >
> > This is strange.  libmpifort is a different version than libmpicxx.so ?
> >
> > If specifying these additional options (manually) get a functional
> > superlu_dist build - you can also specify them to petsc configure via
> >
> > --download-superlu_dist-cmake-arguments=string
> >
> > [but yeah - good to know if this issue persists with petsc/main]
> >
> > Satish
> >
> >
> > On Mon, 28 Mar 2022, Barry Smith wrote:
> >
> > >
> > >   Could you please try with the main branch of PETSc?
> > >
> > >   We've seen similar problems that have at least partially been dealt
> > with in the main branch.
> > >
> > >    Barry
> > >
> > >
> > >
> > > > On Mar 28, 2022, at 4:42 PM, Alexander Lindsay <
> > alexlindsay239 at gmail.com> wrote:
> > > >
> > > > Attached is my configure.log. Error is:
> > > >
> > > >  Could NOT find MPI_C (missing: MPI_C_HEADER_DIR) (found version "4.0")
> > > >
> > > > What's interesting is that cmake does successfully find MPI_CXX and
> > MPI_Fortran albeit in a place I'd rather it not find it (a gcc mpi build
> > located at $HOME/mpich/installed/lib whereas mpicc -show yields what it
> > should at $HOME/mpich/installed-clang/lib).
> > > >
> > > > My same configure line worked well with PETSc 3.15.x. I noticed that
> > these arguments are empty:
> > > >
> > > > -DMPI_C_INCLUDE_PATH:STRING="" -DMPI_C_HEADER_DIR:STRING=""
> > -DMPI_C_LIBRARIES:STRING=""
> > > >
> > > > If I fill those with the correct values and manually cmake
> > superlu_dist, then configure is successful
> > > > <configure.log>
> > >
> >
> >
>